Water Policy Research

HIGHLIGHT

Download this highlight from h�p://iwmi-tata.blogspot.in

Angèle Cauchois

MEASURING THE INVISIBLE Exploi�ng the water-energynexus to es�mate private, urban groundwater dra�

Pumping of groundwater for domes�c uses

is common across the country. Almost

every high-rise, every new housing society

and independent houses have a private

tubewell that services domes�c water

requirements. However, none of these

private tubewells are registered. As a result,

governance ins�tu�ons and policy makers

have no idea about the extent of private

groundwater extrac�on. However, as soon

as the owners acquire an electricity

connec�on, they create a surrogate

registra�on which can be used to quan�fy

the extent of private groundwater dra�.

The objec�ve of this study is to develop a

methodology to es�mate urban

groundwater withdrawal using data from

electricity u�li�es. We argue that the

water-energy nexus can inform

groundwater management and governance

policies in urban India by producing a rapid,

context-sensi�ve and fairly robust es�mate

of private groundwater dra�. Applying our

methodology to selected regions in Gujarat,

we find that private groundwater dra�

represents a significant share of total water

consump�on in these regions.

052 0 1 6

Water Policy Research Highlight-05

1

1. THE INVISIBLE RESOURCE

Groundwater plays a vital role in urban water systems. Most

municipali�es in the world rely on groundwater; private users

too do not hesitate to turn to wells when municipal supply

proves unreliable. In India, the number of privately-owned

urban wells has soared in the past decades. While low

income households rely on tankers, public stand posts and

common dug wells, middle and high income households are

more likely to have the financial capacity to drill their own

bore wells. It has been established that private bore wells are

an important source of domes�c water in several ci�es

(Anand et al. 2005; Raju et al. 2008; Daga 2003) and may

account for the persistent groundwater stress these ci�es

experience. The stage of groundwater development has

reached 864 per cent in Hyderabad, 406 per cent in Chennai,

232 per cent in Gurgaon, 142 per cent in Bangalore, 138 per

cent in Delhi and 102 per cent in Ahmedabad (CGWB 2011a;

CGWB, 2011b).

Most municipali�es fail to acknowledge the important role

that groundwater plays in mee�ng urban water demand

(Narain 2012) and/or the risks associated with groundwater

over extrac�on (Table 1). In this context, the Chennai

Municipality stands out: it enacted its own groundwater

conserva�on act as early as 1987. Since then, external

pressure from the Supreme Court (in the case of Gurgaon) or

from state governments (such as Karnataka or West Bengal)

has pushed certain metropolises to take ac�on to curb

private groundwater dra�, via the issuance of permits for

new abstrac�on structures and the registra�on of exis�ng

wells. None of these regula�ons however, has brought about

expected results (Narain 2012; Grönwall 2013). In her study

of groundwater regula�on in Bangalore, Grönwall (2013)

iden�fies the total absence of data as a major impediment to

effec�ve groundwater regula�on and governance.

Quan�fica�on and monitoring of private groundwater dra�

is both important and difficult. It is important in order to: [a]

manage and respond to groundwater deple�on, which

plagues many Indian towns and ci�es; [b] assess water

demand accurately at the city-scale; and [c] assess

wastewater produc�on and design appropriate wastewater

systems. It is difficult since groundwater is an invisible

resource. Its domes�c exploita�on has reached a

tremendous scale but this is done within the private space of

the residen�al premise, which conceals it from the inquisi�ve

eyes of the regulator. Effec�ve monitoring requires full

coopera�on on the part of the users and past experience has

shown that this is not a given, be it in Mexico, Spain, or India

MEASURING THE INVISIBLE Exploi�ng the water-energy nexus to es�mate private, urban groundwater dra�*

Research highlight based on Cauchois (2015).

RISKS CAUSES CONSEQUENCES

Deple�on Over-exploita�on - Drying up of exis�ng tube wells;- High failing rate of new tube wells; - Increasing depth of the water table implying extra drilling and

pumping costs;- Importa�on of water at high costs from the hinterlands

Land subsidence Over-exploita�on - Lowering of the surface;- Cracking, �l�ng, sinkholes;- Increased vulnerability to floods

Floods Land subsidence - Adverse health effects;- Contamina�on of water;- Damaging of the built environment

Sea water intrusion Over exploita�on in coastal areas - Long-las�ng altera�on of water quality

Pollu�on by contamina�ng agents (fluoride, arsenic, nitrate etc.)

- Over exploita�on - Long-las�ng altera�on of water quality; - Adverse health effects, such as gastro-intes�nal illnesses, dental

and skeletal fluorosis, cancers, poisoning

Table 1: Risks associated with groundwater over exploita�on in urban areas

*This Highlight is based on research carried out under the IWMI-Tata Program (ITP) with addi�onal support from the CGIAR Research Program on Water, Land and Ecosystems (WLE). It is not externally peer-reviewed and the views expressed are of the author/s alone and not of ITP or its funding partners.

2

(Shah 2009). Exis�ng a�empts to assess private groundwater

usage in Indian ci�es are at best vague, and at worst a failure

(CGWB 2011a; Grönwall 2013). At the municipal level, li�le

is known about the actual number, let alone the average

dra�, of private wells and users.

The Central Ground Water Board (CGWB) does publish

groundwater scenarios on a regular basis. However, this

informa�on is available at the district or block level only,

except for twenty-eight urban centers for which

groundwater dra� is computed at the municipal level also.

For domes�c and industrial groundwater use, the following

method is employed:

“The most commonly used method for computa�on of

irriga�on dra� is – number of abstrac�on structures

mul�plied by the unit seasonal dra�. […] Ground water

dra� for domes�c and industrial needs is computed

using the unit dra� method and [is] based on

consump�ve use pa�ern of the popula�on.”

– (CGWB 2011a, p7)

In other words, CGWB's numbers are computed on

popula�on, per capita consump�on, and assump�ons

regarding the number of abstrac�on structures. This method

comes with serious inaccuracies. As we have seen, the

number of abstrac�on structures is a big ques�on mark in

ci�es. In addi�on, the CGWB relies on official standards to

assess per capita consump�on, which bear li�le resemblance

to reality and vary largely across ci�es.

Other methods too have similar limita�ons, either because

they lack accuracy when producing data or because they are

not feasible on a large scale. Narain (2012) advises to deduce

groundwater use based on the quan�ty of sewage a city

produces. However, reliable data on sewage genera�on is

hardly ever available. Raju et al. (2008) and Anand et al.

(2005) try to assess the number of private wells through

systema�c survey. However, these methodologies involve

significant human capital and �me if it is to be replicated in

bigger areas. In addi�on, it gives li�le tangible informa�on as

to the average dra� of one domes�c borewell.

The present study aims at filling this knowledge gap by

proposing a methodology that is able to produce fairly

accurate results with minimal effort, so as to be useful to

policy makers.

2. ESTIMATING GROUNDWATER DRAFT RAPIDLY AND

ACCURATELY

In order to assess domes�c groundwater use, we use

electricity consump�on as a surrogate. The logic behind the

this methodology relies on two basic observa�ons: [a]

domes�c tube wells, in the formal part of the city, use

electric pumps to li� water from the ground; [b] it is possible

to segregate electricity consump�on of domes�c pumps,

either by relying on specific water works (WW) tariffs (in

Gujarat), or by looking at the contracted load of a domes�c

Water Policy Research Highlight-05

connec�on if there is no specific water works tariffs.

Electricity tariffs are set at the distribu�on u�lity level and

differ from one state to the other. In Gujarat, because

electricity u�li�es issue specific tariffs for domes�c premises

and for public or private water works, it becomes rela�vely

easy to compute urban groundwater dra�.

2.1 How can electricity consump�on inform us about

groundwater dra�?

To withdraw groundwater, we need amount of energy β

within a given period of �me. We can establish a simple

correla�on between electricity consump�on and domes�c

groundwater abstrac�on, which illustrates the nexus

between water and energy:

Groundwater withdrawal (L)=β ×Energy consumed (kWh)

The ' ' factor is not a constant; it varies across pumping β

systems. Pumping systems vary in size, age, state, efficiency,

and depth of the well. Depending on these factors, the above

correla�on varies accordingly. There might be as many

different values for as there are pumps in Anand. Engineering β

science teaches us that the different factors impac�ng for the

value of can be captured using the following equa�on:β

x x ⁶Q=

E(kWh) 3.6 10

9.81 h x

Where:

Q = Yearly dra� (L)

h = differen�al head (m) = depth of the bore well + height of

the building/reservoir

E(kWh) = Energy consumed yearly (kWh)

Hence, to calculate the total groundwater dra� of a city, we

need to know the values of the three variables listed below:

§ Units consumed (kWh) by the totality of the water works

connec�ons over a year.

§ Average height of buildings/reservoirs in the city. For

private connec�ons, one can either perform a

representa�ve sample survey of buildings' height in the

city that is being considered (as we proceed in ) or Anand¹

make an assump�on, based on the characteris�cs of the

city. In flat, low-density ci�es such as Mehsana or Anand,

we assume an average of 15 meters. In more ver�cal

ci�es such as Ahmedabad, we take 30 meters as the

average height.

§ Depth of the bore well. The depth depends greatly upon

the city in considera�on. It can be collected either from

households themselves, or by asking drilling companies,

or by using data from the Minor Irriga�on Census. For

Anand, we approached households to determine the

average depth of borewells. For other es�mates, we used

the fourth Minor Irriga�on Census, which produces data

on the depth of irriga�on wells at the district level.

¹We computed the average height of a building by mul�plying its number of floors+1 by 3.1 meters.

3

Water Policy Research Highlight-05

The above equa�on may not produce an accurate es�mate

of groundwater dra� for two reasons. First, in certain

residen�al socie�es and premises, the electricity

consump�on of the buildings' elevator and other public

ligh�ng fixtures might be included in the water works

electricity bill and this may lead to an over-es�ma�on of

groundwater dra�. Second, the equa�on assumes that

groundwater pumps are 100 per cent efficient, which is not

true. In order to address the first issue, we carried out a

survey in the city of Anand to determine whether this was a

common prac�ce and to es�mate the average electricity

consump�on of an elevator. We found that about 13 per

cent of the total waterworks connec�ons have the elevators'

consump�on included in the electricity bill and that the

average annual consump�on is 3,331 kWh. To address the

second issue, we also es�mated the average efficiency of

urban water pumps in Anand to be about 39 per cent. Both

these findings were then incorporated in our computa�on of

annual groundwater dra�.

2.2 An Excel tool to facilitate municipal diagnosis

Overall, this methodology has the advantage of producing a

first-cut es�mate with minimal effort, since a trip to the local

electricity u�lity shall suffice to produce fairly accurate

results. Further, we can facilitate data collec�on and analysis

by designing an easy-to-use excel model. Once the user has

entered relevant informa�on regarding the city being

surveyed, the total private groundwater dra� as well as the

share of private groundwater in total supply is automa�cally

computed (see Table ). 2

3. GROUNDWATER DEPENDENCE IN NORTHERN AND

CENTRAL GUJARAT

The data on which our results are based was collected from

the electricity u�li�es for central and north Gujarat: MGVCL

( Company Ltd.) and UGVCL (Madhya Gujarat Vij U�ar Gujarat

Vij Company Ltd.). The two u�li�es provided us with data at

the circle-level, whose geographical boundaries correspond

roughly to those of eleven administra�ve districts of central

and northern Gujarat . Water orks Gram Panchayats ² w

(WWGP) connec�ons were ignored. Urban popula�on was

determined at the district level based on 2011 Census data.

We assume that in ci�es with popula�on less than 5 , the lakh

inhabitants rely exclusively on groundwater. We applied this

model and associated assump�ons to municipal supply to

test its robustness and then applied it at the district level to ³

obtain regional es�mates for private groundwater dra� (see

Table ). Official es�mates put municipal dra� in Ahmedabad 3

and at 10,950 and 5,913 ML/year respec�vely. Vadodara

Thus, using our methodology produces fairly accurate

es�mates of municipal groundwater dra�.

According to a 2004 survey of several Indian ci�es (Shaban

and Sharma 2007), the average consump�on of an Indian

household is 92 LPCD. How to explain the discrepancies

observed in our results across districts regarding the average

LPCD? Ahmedabad, Mehsana/Patan districts and

Panchmahal/Dahod districts showcase remarkably high

LPCD (111, 121 and 130 respec�vely). Incidentally, they also

boast the deepest bore wells of the en�re region studied

(119, 132 and 105 meters respec�vely). In such water-scarce

areas, using the Minor Irriga�on Census to assess the

average depth of borewells is likely to lead to inaccuracies.

The Minor Irriga�on Census has but only one imprecise

category for very deep wells, namely deeper than 150

meters. This category does not indicate whether the wells

DATA

Average li� consump�on yearly (kwh) 3,331

Private pumps' efficiency (%) 39%

Average number of combined bills (li� + pumps) in WWSP connec�ons (%)

13%

CITY CHARACTERISTICS

Electricity Data

Private units of consump�on yearly (kwh)

Number of private connec�ons

City Characteris�cs

Popula�on (source: 2011 census)

Average depth of borewells (Source: Minor Irriga�on Census)

Average height of buildings (Source: Assump�on)

Total municipal supply from surface water (Source: Municipality)

Total municipal supply from groundwater (Source: Municipality)

RESULTS

Total private groundwater dra�

Share of groundwater in total supply (%)

Share of private groundwater dra� in total supply (%)

Liters per capita per day (LPCD)

Table 2: Model for es�ma�ng groundwater dra�, using Anand as a

benchmark (all units are yearly; water quan�ty is expressed in million

liters unless specified otherwise)

²Namely Ahmedabad, Vadodara, Banaskantha, Sabarkantha, Panch Mahal, Dohad, Anand, Mahesana, Gandhinagar, Patan and Kheda districts. ³The share of units dedicated to drainage, pressure and groundwater extrac�on respec�vely are determined based on Anand municipal data regarding the total pumps' capacity (HP). Technically speaking, equa�ng load to consump�on is a rather eccentric assump�on, yet it gives a rela�vely sa�sfactory es�mate of municipal groundwater dra�. We found that about 56 per cent of electricity consumed by municipal waterworks is dedicated exclusively to groundwater dra�. Other electricity units consumed are dedicated to drainage or pressure pumps. Our results proved to be quite similar to municipal es�mates, which comfort the usability of our methodology.

4

Water Policy Research Highlight-05

concerned are 160 meters on average (which is our

assump�on in this study) or closer to 400 meters deep. As a

result, we may have underes�mated the average depth of

borewells in the districts concerned, leading to an

overes�ma�on of groundwater dra� in the water-scarce

areas of Gujarat. A rapid survey on the ground may resolve

this problem. In the case of populous Ahmedabad, where

groundwater plays but a minor role with respect to total

water supply, this inaccuracy does not introduce much bias

in our results; increasing the average depth of borewells from

the current 119 meters to a hypothe�cal 300 meters will

bring down the LPCD by eight liters only.

In other districts, the average LPCD is lower than Shaban

and Sharma (2007)'s es�mate. This is coherent, given that

the methodology excludes certain categories of actors. The

present results are an under-es�mate of actual private

groundwater dra�. Our star�ng assump�ons, if inaccurate,

are all the more likely to affect our results, as Central and

Northern Gujarat, with the notable excep�on of Ahmedabad

and Vadodara, consist of a constella�on of small towns and

ci�es; light inaccuracies are likely to be amplified by the low

density that characterizes most of the urban areas under

scru�ny.

We observe a loose correla�on between the size of the

major city and private reliance on groundwater, which

confirms findings from other studies (Patel and Krishnan

2009). In Ahmedabad and Vadodara, where both the

municipali�es have almost completely moved away from

District(s)Ahmedabad – Gandhinagar

Vadodara Anand KhedaMehsana,

PatanPanchmahal,

DahodBanaskantha Sabarkantha

TOTAL

Major City(Popula�on)

Ahmedabad (6,352,254)

Vadodara (1,817,191)

Anand (286,921)

Nadiad (225,071)

Mehsana (190,189)

Godhra (143,644)

Palanpur (140,344)

Himmatnagar (81,137)

Surface Water (ML)

268,279(CGWB 2011a)

53,217(Narain 2012)

0 0 0 0 0 0 321,496

Municipal Groundwater Dra� (ML)

10,776 5,959 9,149 8,584 8,799 10,067 25,001 10,061 88,398

Private Groundwater Dra� (ML)

49,199 8,880 11,899 6,787 15,237 8,582 7,395 2,442 110,422

Share GW/Total Supply (%)

18% 22% 100% 100% 100% 100% 100% 100% 38%

Share Private GW Dra�/Total Supply (%)

15% 13% 57% 44% 63% 46% 23% 20% 21%

LPCD (L) 111 88 79 67 121 81 130 88 96

Table 3: Urban groundwater dra� in North and Central Gujarat

groundwater sources while expanding the municipal water

supply network well beyond na�onal averages, private

groundwater dra� does not contribute to a large extent in

the total water supply. Reliance on groundwater resources is

found to be the highest in districts coun�ng medium-sized

and growing ci�es such as Anand, Mehsana, Godhra or

Nadiad. The low share of private groundwater dra� in

Banaskantha (23 per cent) and Sabarkantha (20 per cent)

may also be correlated to city size. In both districts, urban

areas are like a collec�on of very small towns of less than

50,000 inhabitants, almost skin to villages. While further

research is needed to validate this hypothesis, it might well

be the case that, similar to rural areas, residents have not

equipped themselves with electric pumps within their

premise, especially in these water-scarce areas where drilling

a private well represents a substan�al investment.

4. ESTIMATING GROUNDWATER DRAFT OUTSIDE GUJARAT

In other states, there are no separate connec�ons and tariffs

for water works connec�ons. However, domes�c pumps

would s�ll stand out in the official data due to their

contracted load, which is higher than other domes�c

connec�ons. Garg et al. (2014) studied the load profile of

Gujara� households for various income groups (Figure 1).

According to their findings, even AC-equipped high income

households do not use more than 2 kW of power at any

given point of �me. These findings are consistent with our

own research. According to officials' personal es�mates for

the ci�es of Jaipur, Jaisalmer, Jodhpur, Kota, Anand and

5

Water Policy Research Highlight-05

Hyderabad, households contract 3 kW of load on average for

domes�c purposes; 6 kW in Bangalore; between 3 and 5 kW

in Delhi; up to 10 kW in Chennai. These are adventurous and

quick personal es�mates from officials; according to

secondary data collected for the district of Vadodara, urban

domes�c connec�ons score an average connected load of

1.31 kW. By contrast, domes�c pumps (WW.PR connec�ons)

in Gujarat show much higher contracted load on average:

Figure 1: Load profile for income categories during (a) summer and (b) winter.

7 kW in water-abundant Anand City, 22 kW in Ahmedabad

district, and 10 kW in Vadodara district. This average reaches

a dazzling 34 kW in Mehsana. It should be noted that these

domes�c connec�ons o�en service an en�re society, colony

or high-rise apartment building with several inhabitants, and

not individual household connec�ons. Therefore, it is

possible to segregate domes�c connec�ons from water

works connec�ons based on their load profile.

[HI: High Income; HMI: Higher Middle Income; MMI: Middle Medium Income; LMI: Lower Medium Income; ULI: Upper Low Income]

Source: Garg et al. 2014

0

0.3

0.6

0.9

1.2

1.5

1 3 5 7 9 11 13 15 17 19 21 23

kW/H

ousehold

Hour

Summer

HI HMI MMI LMI ULI

a

0

0.3

0.6

0.9

1.2

1.5

1 3 5 7 9 11 13 15 17 19 21 23

kW/H

ousehold

Hour

Winter

HI HMI MMI LMI ULI

b

6

Water Policy Research Highlight-05

REFERENCES

Anand, K.D., Raju, K.V., Praveen, N., Deepa, N., and Latha N. (2005): “Urban water supplies dependency on groundwater: Can it sustain? – Ini�al

Findings from South Indian ci�es”. Unpublished paper, Anand: IWMI-Tata Water Policy Program.

Cauchois, A. (2015): “Es�ma�ng urban groundwater dra� in central and northern Gujarat using electricity consump�on as a surrogate: A

Methodology”. Unpublished internship report, Anand: IWMI-Tata Water Policy Program.

Central Ground Water Board (CGWB). (2011a): “Dynamic ground water resources of India” (as on 31 March 2009). Retrieved from:

h�p://www.cgwb.gov.in/GW_assessment.html

Central Groundwater Board (CGWB). (2011b): “Groundwater Scenario in Major Ci�es of India”. Retrieved from: www.cgwb.gov.in

Daga, S. (2003): “Private supply of water in Delhi”. Discussion Paper, Centre for Civil Society. Retrieved from:

h�p://ccs.in/internship_papers/2003/chap18.pdf

Garg, A., Shukla, P.R., Maheshwari, J., and Upadhyay, J. (2014): “An assessment of household electricity load curves and corresponding CO2

marginal abatement cost curves for Gujarat state, India. Energy Policy, 66: 568-584

Grönwall, J.T. (2013): “Groundwater Governance in India: Stumbling Blocks for Law and Compliance”. WGF Report No. 3, Stockholm: Stockholm

Interna�onal Water Ins�tute (SIWI). Retrieved from:

h�p://www.watergovernance.org/documents/WGF/Reports/2013_No3_Groundwater_Governance_India-3-new.pdf

Narain, S. (2012): “Excreta ma�ers: How urban India is soaking up water, pollu�ng rivers and drowning in its own waste”. Vol. I & II. New Delhi:

Center for Science and Environment.

Patel, A. and Krishnan, S. (2009): “Groundwater situa�on in urban India: Overview, opportuni�es and challenges. In Amarasinghe, Upali A.; Shah,

Tushaar; Malik, R. P. S. (Eds.). Strategic Analyses of the Na�onal River Linking Project (NRLP) of India, Series 1: India's water future: scenarios and

issues. Colombo, Sri Lanka: Interna�onal Water Management Ins�tute (IWMI) pp.367-380.

Raju, K.V., Manasi S., and Latha N. (2008): “Emerging groundwater crisis in urban areas: A case study of Ward No. 39, Bangalore: Ins�tute for Social

and Economic Change, Working paper 196. Retrieved from: h�p://www.isec.ac.in/WP%20-%20196.pdf

Shaban, A., and Sharma, R.N. (2007): “Water Consump�on Pa�erns in Domes�c Households”. Economic & Poli�cal Weekly, 42(23): 2190 – 2197

Shah, T. (2009): “Taming the anarchy: Groundwater governance in South Asia”. Washington, DC: RFF Press

5. CONCLUSION

In the present study, we used the water-energy nexus to

compute private, municipal groundwater dra� in Central and

Northern Gujarat. It produced compelling evidence that

private groundwater dra� represents a significant share of

total water consump�on in these regions. Since exis�ng

methodologies have failed to produce very accurate data in

this regard, this methodology is a first step towards a greater

understanding of groundwater pa�erns of use in urban

contexts. Using the water-energy nexus to compute

groundwater dra� offers local policy makers the capacity to

measure rapidly, effec�vely and fairly accurately at the city

level.

IWMI Headquarters127 Sunil MawathaPelawa�e, Ba�aramullaColombo, Sri LankaMailing AddressP. O. Box 2075, Colombo, Sri LankaTel: +94 11 2880000, 2784080 Fax: +94 11 2786854 Email: iwmi@cgiar.org Website: www.iwmi.org

IWMI-Tata Water Policy Program“Jal Tarang”Near Smru� Apartments, Behind IRMA GateMangalpura, Anand 388001, Gujarat, IndiaTel: +91 2692 263816, 263817 Email: iwmi-tata@cgiar.org

IWMI is amember ofthe CGIARConsortiumand leadsthe:

RES EARCH PROGRA ON

Water, Land and Ecos stems

About the IWMI-Tata Program and Water Policy Highlights

The IWMI-Tata Water Policy Program (ITP) was launched in

2000 as a co-equal partnership between the Interna�onal

Water Management Ins�tute (IWMI), Colombo and Sir Ratan

Tata Trust (SRTT), Mumbai. The program presents new

perspec�ves and prac�cal solu�ons derived from the wealth

of research done in India on water resource management. Its

objec�ve is to help policy makers at the central, state and local

levels address their water challenges – in areas such as

sustainable groundwater management, water scarcity, and

rural poverty – by transla�ng research findings into prac�cal

policy recommenda�ons. Through this program, IWMI

collaborates with a range of partners across India to iden�fy,

analyze and document relevant water management

approaches and current prac�ces. These prac�ces are

assessed and synthesized for maximum policy impact in the

series on Water Policy Highlights and IWMI-Tata Comments.

Water Policy Highlights are pre-publica�on discussion papers

developed primarily as the basis for discussion during ITP's

Annual Partners' Meet. The research underlying these

Highlights was funded with support from Interna�onal Water

Management Ins�tute (IWMI), Tata Trusts, CGIAR Research

Program on Water, Land and Ecosystems (WLE) and CGIAR

Research Program on Climate Change, Agriculture and Food

Security (CCAFS). However, the Highlights are not externally

peer-reviewed and the views expressed are of the author/s

alone and not of ITP or any of its funding partners.